Justin Jincaid

Aident AI Beta 2 - Open-world automations, managed in plain English

Meet the future of work—AI that actually runs with you. Build and manage open-world automations in plain English across Discord, Slack, X, Shopify, and more with 1000+ integrations, 23000+ actions, and 1000+ templates. Trigger on real-world events, get updates in your favorite chat apps via IM + MCP, and monitor runs, approvals, and issues from one live dashboard.

Add a comment

Replies

Best
Edward.J

Hi everyone👋. I’m Edward, the designer of Aident AI

One thing we kept running into while building automation tools: the hardest part isn’t wiring steps together — it’s handling the messy, unpredictable real world.

For Beta 2, I tried to rethink automation from a design perspective:

1. From UI design → Instruction design.
Instead of drawing flows, users describe intent. The system interprets and structures it.
That shift changed how we design everything.

2. Reducing cognitive overhead.
Node graphs are powerful — but they demand constant mental mapping.
We asked: can automation feel more like delegating to an assistant than managing a dashboard?

3. Closing the “demo vs real work” gap.
Supporting live triggers (Discord, Slack, Shopify, etc.) forced us to design for reliability, not just flexibility.

Also curious about this: for non-technical users, what’s been the hardest part about turning a vague goal into something automation can actually run? 👇

Would love to hear real frustrations — they help us design better.

Andrey Chernyshev

@edwardjia Hey Edward what is the exact moment or event in Aident that makes me think: "This is worth paying for"?
I mean something like aha moment when I finally will prove the value?

Edward.J

@andrey_chernyshev1 Hey Andrey. Good question. For me the “aha moment” is when an automation reaches a decision point and the agent asks you instead of blindly continuing. You’ll see it show up as an “Awaiting” state in the dashboard and realize it’s not a black box.

Another moment is how quickly you can go from building a playbook to actually using it. You can define it in natural language and trigger it from apps like messaging tools, so the agent becomes something you interact with during work.

Our thinking is that AI is incredibly efficient as a tool, but it’s not always the decision maker. Ideally that’s when it clicks: the automation is running with you, not just in the background.

Shake Lyu

@edwardjia How does Aident AI help users identify and manage these unpredictable edge cases if they don't even know they need to account for them yet?"

Pashupathi Mali

@edwardjia One thing I've noticed while building around AI-powered systems is that the biggest challenge isn’t describing the happy path, it's dealing with uncertainty when the underlying tools behave inconsistently.

Once automations start relying on external APIs or models, small reliability issues cascade into confusing failures. Curious how Aident handles situations where one of the connected tools behaves unexpectedly mid-run.

Joao Seabra

"AI was smart, the glue was not" is the most honest description of why the current automation stack keeps breaking. Zapier plus RPA plus prompt engineering duct-taped together fails in exactly the ways you'd expect and the 1000th time is never the last time.

The shift from UI design to instruction design that Edward describes is the right framing for what makes this feel different. Node graphs are powerful but they're a UX paradigm built for engineers. Describing intent to a teammate and having it figure out the structure is a completely different contract with the user.

To answer Edward's question from someone building their own AI platform: the hardest part of turning a vague goal into a running automation is usually defining the edge cases. People describe the happy path clearly, but when something unexpected happens mid-run, the automation either fails silently or does something wrong. How does Aident handle ambiguous situations mid-playbook? Does it stop and ask, or make a judgment call? Congrats on the launch! 🤖

Yulei Sheng

@joao_seabra Thanks! Let us know if you have any feedback.

Edward.J

@joao_seabra Thanks for your comment. Really appreciate this perspective “AI was smart, the glue was not” captures the current automation stack problem.

The edge case issue you mentioned is exactly where most automations break. People describe the happy path well, but the moment something unexpected happens mid-run, traditional workflows either fail silently or do the wrong thing.

In Aident we tried to design around that reality rather than assuming everything can be predefined.

When a playbook hits an ambiguous situation, the agent doesn’t just guess blindly. It can pause and surface that moment as an “Awaiting” state in the dashboard. From there, the user can jump in and interact directly with the agent — provide input, approve a decision, or modify the step if needed.

The idea is that automation shouldn’t be purely autonomous or purely manual, it should feel more like collaborating with an assistant that knows when to ask for help.

Saira Wang

Does Aident sit on top of existing tools or replace them?

I already have a bunch of Zaps running, so migration cost is a real concern for me.

Yulei Sheng

@saira_wang if you can email me(yulei.sheng@aident.ai) a few Zaps examples, i can help you to migrate

Kimi Lu

@saira_wang it should be too hard. you can export all your zaps into one json, and giving it to our agent and our agent can turn each of the zaps into equivalent playbook for the automation, almost identical.

Kimi Lu

Hey guys, I'm Kimi, founder of Aident - quite excited to share a new version of Aident, the Beta 2!

I built this product because of my previous bad experience of my old automation setup, which now sounds like a caveman: gluing Zapier with RPA to prompt ChatGPT and other websites. When it broke the 1000th time, I decided that something needs to change. AI was smart... yet the glue was not.

Now, I am building Aident with my team to make it crazy easy to turn your life to be AI-native with almost zero frustration: you describe what you want like talking to a teammate and Aident as an Executive Assistant would turn that into a playbook that runs and delivers.

Being able to glue anything you already love is absolutely important and fundamental to automate anything. And, that's why we worked hard to expand our built-in integrations to 1000+ with 23k+ actions available. This would cover almost anything around your work and your life.

Our next step is to continue to expand it to even one more scale up, e.g. 10k+ integrations and absolutely 10x more skills in the next version. Stay tuned!

Here is Aident Beta 2, and yes, it is still beta, so if it does something dumb, please tell me. But, if it clicks for you, I'd really appreciate some love and an upvote! 🙏🙏🙏

Ask me anything - I'll be in the comments all day!

- Kimi

Luke Guo

Hello everyone, this is Luke, Eng @ Aident AI. I'm super happy that we are launching Aident Beta 2 today!

I've personally struggled with how a non-tech person would adopt AI in their daily workflows. Many people around me wanted to use AI to organize Google Sheets, review emails, or coordinate tasks in Slack, but the reality was that most tools still required too much technical setup or rigid workflow building.

While working on Aident, one thing we kept focusing on was making automation feel less like programming a machine, and more like delegating work to a teammate. Instead of designing complicated node graphs, you can simply describe what you want in plain English, and Aiden turns it into a playbook that actually runs across your tools.

We’re still very early and actively learning from users. If you try Aident Beta 2 and something breaks, feels confusing, or you just have ideas on how it should work, please let us know. Feedback like that directly shapes what we build next.

Really excited to hear what kinds of workflows you’d want AI to run for you!

Gabriel Abi Ramia

Really cool concept — automations described in plain English is where everything is heading. We built something similar for our AI routing at TubeSpark (AI for YouTube creators): each task type is described in natural language and the system picks the best AI model automatically. Curious — how do you handle edge cases where the plain English instruction is ambiguous? That's been our biggest challenge with natural language interfaces.

Luke Guo

@aitubespark Thank you! When instructions are ambiguous, our agent will ask clarification questions and provide proactive solution options for user to choose from. So Aident will only start generating automations when it understands the intent clearly.

Yulei Sheng

Hi friends, I'm Yulei from Aident AI team! I'm beyond excited to share beta 2 with ya'll.

We made multiple 10x improvements compared to beta 1. Just to name a few features that i personally use daily:

  1. time & event triggers! i have workflows that run at certain time everyday and run when conditions met.

  2. discord & telegram! sometime, i'm on the go and just chat with Aiden in discord & telegram

  3. Agent Dashboard, the is the future of work, i manage a bunch of agents here, i can see results, unblock agents, and be a good manager here!

We are still super early, would love you to try Beta 2 and share feedback!

Justified Wang

Huge congrats on the launch! I've been looking for a way to bridge my Slack and Google Sheets without the node-graph headache. Signing up for Beta 2 now, and I'll be sure to send over some honest feedback! 🚀

Kimi Lu

@justified_wang great! would really appreciate your feedbacks!

Janice

As a product manager with no coding experience, I’m always looking for tools that can turn ideas into actionable workflows. Aident AI looks very promising, and I’m excited to try it to see how it can simplify processes and boost productivity.

Yulei Sheng

@janicelewis00 Give it a try and let us know!

Kimi Lu

@janicelewis00 perfect! happy to hear your feedbacks!

Bill Chirico

How does this differ from OpenClaw?

Yulei Sheng

@billchirico i personally use both but i have bias, by having workflows written in docs(agent wrote those docs based on my input), i have higher confidence and know the agent deliverables are more likely to be consistent, this become more important when my tasks get complicated.

123
Next
Last